An object-oriented Bayesian network model for the quantitative risk assessment of navigational accidents in ice-covered Arctic waters
Shanshan Fu,Yue Zhang,Mingyang Zhang,Bing Han,Zhongdai Wu
DOI: https://doi.org/10.1016/j.ress.2023.109459
2023-06-23
Abstract:Merchant ship operations in the ice-covered Arctic waters may encounter traditional navigational accident risks (i.e., grounding, collision, etc.) and risks from sea ice, such as ship besetting in ice. However, describing, modeling, and quantifying the multiple risks in ice navigation are challenges from maritime risk assessment perspective. This paper proposes an object-oriented Bayesian network (OOBN) model for the quantitative risk assessment of multiple navigational accidents in ice-covered Arctic waters. The OOBN model makes use of the accident database from Lloyd's intelligence and maritime accident investigation reports. The proposed model decomposes navigational accidents into five levels based on accident causation theory: environment, unsafe condition, unsafe act, probability of navigational accident, and consequence of the navigational accident. Consequently, collision, grounding, ship besetting in ice, and ship–ice collision accidents are selected as the cases to interpret the quantitative risk assessment for navigational risk factors identification, risk analysis, and evaluation. The results demonstrate that (1) the risk is the highest in grounding accidents, followed by besetting in ice, collision, and ship–ice collision in ice-covered Arctic waters; (2) unsafe speed and unsafe condition are the critical mutual factors of these four accident scenarios; (3) and the critical risk influencing factors for the specific navigational accidents are identified to propose corresponding risk control options. The proposed OOBN model can be used for quantitative risk assessment of navigational accidents in ice-covered Arctic waters.
engineering, industrial,operations research & management science